Plan for more VLM training support?

[shafiqabedin]

Is there a plan for more VLM training examples in trl? Right now I only see dpo_vlm and sft_vlm in the example. Has anyone created/played with other techniques for VLMs that's state-of-the-art?

[shirinyamani]

We plan to have the support for Vision models hopefully sometime soon! Cannot give a date unfortunately!